Non-Disruptive CDP/R Service Transition via WWPN Redirection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for transitioning continuous data protection and replication (CDP/R) services between appliance nodes are disruptive to application hosts, expensive, and complex, often requiring dynamic multi-pathing drivers and complex coordination, leading to performance degradation and conflicts.
Innovation Solution
A method involving determining pending transactions, unregistering and registering world wide port names, associating logical unit numbers, and logging ports to transition CDP/R services from one appliance node to another without disruption, using techniques like virtualized port names and extended link services to maintain continuous service.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If CDP/R services are transitioned from one appliance node to another using traditional methods, then service continuity is disrupted, but this causes unacceptable service interruption to application hosts
Solution Approach 1:
The system performs preliminary actions by detecting pending transactions before initiating the transition. The source appliance node identifies and handles outstanding I/O operations in advance, ensuring they are either completed or properly redirected before the WWPN switching occurs, thus preventing service disruption
Solution Approach 2:
The system uses an intermediary mechanism where the source appliance node acts as a temporary mediator to detect and manage pending transactions. This intermediary role allows smooth handover of service responsibilities from source to target appliance node without direct host involvement, maintaining service continuity
2Adaptability or versatility
If Active/Passive configuration with DMP driver is used for failover, then path redundancy is achieved, but device complexity and coordination requirements increase significantly
Solution Approach 1:
The system extracts the complexity of path management and failover coordination from the host node by implementing these functions entirely within the appliance cluster. The host node simply sees WWPN changes without needing DMP drivers or coordination software, eliminating the complexity while maintaining path redundancy
Solution Approach 2:
The system uses virtualization to create a copy of the WWPN identity that can be moved between appliance nodes. Instead of managing multiple physical paths and complex routing, the system copies the logical WWPN identifier from the source node to the target node, simplifying the failover mechanism while maintaining redundancy
3Reliability
If Active/Active configuration with multiple appliance nodes is used, then service availability is improved, but data synchronization complexity and performance degradation occur
Solution Approach 1:
The system performs preliminary detection of pending transactions on the source appliance node before initiating the transition. This advance detection ensures that data synchronization is handled proactively, with the target node receiving and processing outstanding operations before the WWPN switch, eliminating synchronization complexity during failover
Solution Approach 2:
The source appliance node serves as an intermediary that detects and manages pending transactions, acting as a buffer between the host and target appliance node. This intermediary mechanism ensures data consistency without requiring complex distributed locking or synchronization protocols across multiple active nodes
Data Source
AI summary
Techniques for CDP/R services are disclosed. In one particular exemplary embodiment, the techniques may be realized as a method of transitioning continuous data protection and replication comprising determining whether a first appliance node connected to a switched fabric contains one or more transactions received from a host node, unregistering a world wide port name of a target port of the first appliance node, registering the world wide port name to a target port of a second appliance node connected to the switched fabric, associating one or more logical unit numbers of the second appliance node with the target port of the second appliance node, exporting the one or more logical unit numbers of the second appliance node, logging the target port of the second appliance node into a switched fabric, and logging the target port of the second appliance node into a remote node port of the host node.